Correlations and Causation
Kudos to Dr. Morton E. Tavel for his article “Correlations: How Do We Ever Establish Definite Causation?” in SI (September/ October 2015, pp. 54–55). His reference to Hill’s 1965 report by example reinforced grounds for crucial conditions in cause-effect assumptions regarding the linkage between cigarette smoking and cancer. The query from Jim Jackson regarding “correlation is not causation” is raised often in classrooms and scientific critiques. Yet the quotation minimizes the probability that under bounded conditions, as explained in Tavel’s article (e.g., see Hill’s seven guidelines, point 2 “Consistency of association,” p. 55), correlation may be one indicator of “causation.” That is, if all known controls are implemented and if a statistical level of significance (p < .001) is agreed upon, then independent and dependent variables would be both correlated and assumed causally related.